
import pandas as pd 
import numpy as np 
from sklearn import linear_model
from sklearn import metrics
import xgboost as xgb
from xgboost import plot_importance
from sklearn.metrics import explained_variance_score, mean_squared_error, r2_score
import seaborn as sns
from sklearn.preprocessing import StandardScaler
import plotly.express as px
import plotly.offline as pyo
import plotly.graph_objs as go
from sklearn.preprocessing import LabelEncoder
#Hide warnings
import warnings
warnings.filterwarnings('ignore')
import matplotlib.pyplot as plt
# Set plot preference
plt.style.use(style='ggplot')
plt.rcParams['figure.figsize'] = (10, 6)
import seaborn as sns
import matplotlib.pyplot as plt


def_data = pd.read_csv('df_def.csv')
fwd_data = pd.read_csv('df_fwd.csv')
gk_data = pd.read_csv('df_gk.csv')
mid_data = pd.read_csv('df_mid.csv')


def_data.head()


fwd_data.head()


gk_data.head()


mid_data.head()


def_data.info()

<class 'pandas.core.frame.DataFrame'>
RangeIndex: 279 entries, 0 to 278
Data columns (total 47 columns):
 #   Column                  Non-Null Count  Dtype  
---  ------                  --------------  -----  
 0   Unnamed: 0              279 non-null    int64  
 1   Name                    279 non-null    object 
 2   Appearances             278 non-null    float64
 3   Goals                   278 non-null    float64
 4   Wins                    278 non-null    float64
 5   Losses                  278 non-null    float64
 6   Clean sheets            276 non-null    float64
 7   Goals Conceded          276 non-null    float64
 8   Tackles                 278 non-null    float64
 9   Tackle success %        278 non-null    object 
 10  Last man tackles        276 non-null    float64
 11  Blocked shots           278 non-null    float64
 12  Interceptions           278 non-null    float64
 13  Clearances              278 non-null    object 
 14  Headed Clearance        278 non-null    object 
 15  Clearances off line     276 non-null    float64
 16  Recoveries              278 non-null    object 
 17  Duels won               278 non-null    object 
 18  Duels lost              278 non-null    object 
 19  Successful 50/50s       278 non-null    float64
 20  Aerial battles won      278 non-null    object 
 21  Aerial battles lost     278 non-null    float64
 22  Own goals               276 non-null    float64
 23  Errors leading to goal  278 non-null    float64
 24  Assists                 278 non-null    float64
 25  Passes                  278 non-null    object 
 26  Passes per match        278 non-null    float64
 27  Big Chances Created     278 non-null    float64
 28  Crosses                 278 non-null    object 
 29  Cross accuracy %        278 non-null    object 
 30  Through balls           278 non-null    float64
 31  Accurate long balls     278 non-null    object 
 32  Yellow cards            278 non-null    float64
 33  Red cards               278 non-null    float64
 34  Fouls                   278 non-null    float64
 35  Offsides                278 non-null    float64
 36  Headed goals            278 non-null    float64
 37  Goals with right foot   278 non-null    float64
 38  Goals with left foot    278 non-null    float64
 39  Hit woodwork            278 non-null    float64
 40  Big chances missed      2 non-null      float64
 41  Freekicks scored        2 non-null      float64
 42  Goals per match         2 non-null      float64
 43  Penalties scored        2 non-null      float64
 44  Shooting accuracy %     2 non-null      object 
 45  Shots                   2 non-null      float64
 46  Shots on target         2 non-null      float64
dtypes: float64(33), int64(1), object(13)
memory usage: 102.6+ KB


def_data.isnull().sum()

Unnamed: 0                  0
Name                        0
Appearances                 1
Goals                       1
Wins                        1
Losses                      1
Clean sheets                3
Goals Conceded              3
Tackles                     1
Tackle success %            1
Last man tackles            3
Blocked shots               1
Interceptions               1
Clearances                  1
Headed Clearance            1
Clearances off line         3
Recoveries                  1
Duels won                   1
Duels lost                  1
Successful 50/50s           1
Aerial battles won          1
Aerial battles lost         1
Own goals                   3
Errors leading to goal      1
Assists                     1
Passes                      1
Passes per match            1
Big Chances Created         1
Crosses                     1
Cross accuracy %            1
Through balls               1
Accurate long balls         1
Yellow cards                1
Red cards                   1
Fouls                       1
Offsides                    1
Headed goals                1
Goals with right foot       1
Goals with left foot        1
Hit woodwork                1
Big chances missed        277
Freekicks scored          277
Goals per match           277
Penalties scored          277
Shooting accuracy %       277
Shots                     277
Shots on target           277
dtype: int64


def_data = def_data.drop(['Big chances missed','Freekicks scored','Goals per match','Penalties scored','Shooting accuracy %','Shots','Shots on target'],axis=1)
def_data=def_data.dropna(axis=0)


corr=def_data.corr()
# Set up the matplotlib plot configuration
#
f, ax = plt.subplots(figsize=(20, 20))
#
# Generate a mask for upper traingle
#
mask = np.triu(np.ones_like(corr, dtype=bool))
#
# Configure a custom diverging colormap
#
cmap = sns.diverging_palette(230, 20, as_cmap=True)
#
# Draw the heatmap
#
sns.heatmap(corr, annot=True, mask = mask, cmap=cmap)

<AxesSubplot:>


import plotly.express as px
Goals=pd.DataFrame(def_data.groupby(['Name'])['Goals'].sum())
Goals=Goals.sort_values(by=['Goals'])
Goals=Goals.tail(5)
fig = px.bar(Goals, y='Goals',color=Goals.index)
fig.show()


import plotly.express as px
Yellow_cards=pd.DataFrame(def_data.groupby(['Name'])['Yellow cards'].sum())
Yellow_cards=Yellow_cards.sort_values(by=['Yellow cards'])
Yellow_cards=Yellow_cards.tail(5)
fig = px.bar(Yellow_cards, y='Yellow cards',color=Yellow_cards.index)
fig.show()


import plotly.express as px
Red_cards=pd.DataFrame(def_data.groupby(['Name'])['Red cards'].sum())
Red_cards=Red_cards.sort_values(by=['Red cards'])
Red_cards=Red_cards.tail(5)
fig = px.bar(Red_cards, y='Red cards',color=Red_cards.index)
fig.show()


def_data['Tackle success %']=def_data['Tackle success %'].str.replace("%",'')
l=list(def_data['Tackle success %'])
for i in range(len(l)):
    l[i]=int(l[i])
def_data['Tackle success %']=l
Tackle_success=def_data[['Name','Tackle success %']]
Tackle_success=Tackle_success.sort_values(by=['Tackle success %'])
Tackle_success=Tackle_success.tail(15)
fig = px.bar(Tackle_success,x='Name', y='Tackle success %',color='Name')
fig.show()


import re
def_data['Duels won']=def_data['Duels won'].str.replace(',', '')
def_data['Duels won']=def_data['Duels won'].str.replace("'", '')
l=list(def_data['Duels won'])
for i in range(len(l)):
    l[i]=int(l[i])
def_data['Duels won']=l

Duels_wons=pd.DataFrame(def_data.groupby(['Name'])['Duels won'].sum())
Duels_wons=Duels_wons.sort_values(by=['Duels won'])
Duels_wons=Duels_wons.tail(5)
fig = px.bar(Duels_wons, y='Duels won',color=Duels_wons.index)
fig.show()


def_data['Duels lost']=def_data['Duels lost'].str.replace(',', '')
l2=list(def_data['Duels lost'])
for i in range(len(l2)):
    l2[i]=int(l2[i])
def_data['Duels lost']=l2

Duels_lost=pd.DataFrame(def_data.groupby(['Name'])['Duels lost'].sum())
Duels_lost=Duels_lost.sort_values(by=['Duels lost'])
Duels_lost=Duels_lost.tail(5)
fig = px.bar(Duels_lost, y='Duels lost',color=Duels_lost.index)
fig.show()


def_data['Aerial battles won']=def_data['Aerial battles won'].str.replace(',', '')
l3=list(def_data['Aerial battles won'])
for i in range(len(l3)):
    l3[i]=int(l3[i])
def_data['Aerial battles won']=l3

Aerial_battles_won=def_data[['Name','Aerial battles won']]
Aerial_battles_won=Aerial_battles_won.sort_values(by=['Aerial battles won'])
Aerial_battles_won=Aerial_battles_won.tail(5)
fig = px.bar(Aerial_battles_won, x='Name',y='Aerial battles won',color='Name')
fig.show()


l4=list(def_data['Aerial battles lost'])
for i in range(len(l4)):
    l4[i]=int(l4[i])
def_data['Aerial battles lost']=l4

Aerial_battles_lost=def_data[['Name','Aerial battles lost']]
Aerial_battles_lost=Aerial_battles_lost.sort_values(by=['Aerial battles lost'])
Aerial_battles_lost=Aerial_battles_lost.tail(5)
fig = px.bar(Aerial_battles_lost, x='Name',y='Aerial battles lost',color='Name')
fig.show()


l5=list(def_data['Errors leading to goal'])
for i in range(len(l5)):
    l5[i]=int(l5[i])
def_data['Errors leading to goal']=l5

Errors_leading_goal=def_data[['Name','Errors leading to goal']]
Errors_leading_goal=Errors_leading_goal.sort_values(by=['Errors leading to goal'])
Errors_leading_goal=Errors_leading_goal.tail(5)
fig = px.bar(Errors_leading_goal, x='Name',y='Errors leading to goal',color='Name')
fig.show()


import math
l6=list(def_data['Passes per match'])
for i in range(len(l6)):
    l6[i]=math.floor(l6[i])
def_data['Passes per match']=l6

Passes_per_match=def_data[['Name','Passes per match']]
Passes_per_match=Passes_per_match.sort_values(by=['Passes per match'])
Passes_per_match=Passes_per_match.tail(5)
fig = px.bar(Passes_per_match, x='Name',y='Passes per match',color='Name')
fig.show()


Duel_won=def_data[['Name','Duels won']]
Duel_won=Duel_won.sort_values(by=['Duels won'])
Duel_won=Duel_won.tail(5)
fig = px.bar(Duel_won, x='Name',y='Duels won',color='Name')
fig.show()


Duel_lost=def_data[['Name','Duels lost']]
Duel_lost=Duel_lost.sort_values(by=['Duels lost'])
Duel_lost=Duel_lost.tail(5)
fig = px.bar(Duel_lost, x='Name',y='Duels lost',color='Name')
fig.show()


def_data['Cross accuracy %']=def_data['Cross accuracy %'].str.replace("%",'')
l=list(def_data['Cross accuracy %'])
for i in range(len(l)):
    l[i]=int(l[i])
def_data['Cross accuracy %']=l
Cross_accuracy=def_data[['Name','Cross accuracy %']]
Cross_accuracy=Cross_accuracy.sort_values(by=['Cross accuracy %'])
Cross_accuracy=Cross_accuracy.tail(10)
fig = px.bar(Cross_accuracy,x='Name', y='Cross accuracy %',color='Name')
fig.show()


fwd_data.isnull().sum()

Unnamed: 0                  0
Name                        0
Appearances                 0
Goals                       0
Wins                        0
Losses                      0
Goals per match             0
Headed goals                0
Goals with right foot       0
Goals with left foot        0
Penalties scored            0
Freekicks scored            0
Shots                       0
Shots on target             0
Shooting accuracy %         0
Hit woodwork                0
Big chances missed          0
Assists                     0
Passes                      0
Passes per match            0
Big Chances Created         0
Crosses                     0
Yellow cards                0
Red cards                   0
Fouls                       0
Offsides                    0
Tackles                     0
Blocked shots               0
Interceptions               0
Clearances                  0
Headed Clearance            0
Accurate long balls       175
Aerial battles lost       175
Aerial battles won        175
Cross accuracy %          175
Duels lost                175
Duels won                 175
Errors leading to goal    175
Recoveries                175
Successful 50/50s         175
Tackle success %          175
Through balls             175
dtype: int64


fwd_data = fwd_data.drop(['Accurate long balls', 'Aerial battles lost',
       'Aerial battles won', 'Cross accuracy %', 'Duels lost', 'Duels won',
       'Errors leading to goal', 'Recoveries', 'Successful 50/50s',
       'Tackle success %', 'Through balls'],axis=1)
fwd_data=fwd_data.dropna(axis=0)


corr=fwd_data.corr()
# Set up the matplotlib plot configuration
#
f, ax = plt.subplots(figsize=(20, 20))
#
# Generate a mask for upper traingle
#
mask = np.triu(np.ones_like(corr, dtype=bool))
#
# Configure a custom diverging colormap
#
cmap = sns.diverging_palette(230, 20, as_cmap=True)
#
# Draw the heatmap
#
sns.heatmap(corr, annot=True, mask = mask, cmap=cmap)

<AxesSubplot:>


Goals=pd.DataFrame(fwd_data.groupby(['Name'])['Goals'].sum())
Goals=Goals.sort_values(by=['Goals'])
Goals=Goals.tail(5)
fig = px.bar(Goals, y='Goals',color=Goals.index)
fig.show()


Yellow_cards=pd.DataFrame(fwd_data.groupby(['Name'])['Yellow cards'].sum())
Yellow_cards=Yellow_cards.sort_values(by=['Yellow cards'])
Yellow_cards=Yellow_cards.tail(5)
fig = px.bar(Yellow_cards, y='Yellow cards',color=Yellow_cards.index)
fig.show()


Red_cards=pd.DataFrame(fwd_data.groupby(['Name'])['Red cards'].sum())
Red_cards=Red_cards.sort_values(by=['Red cards'])
Red_cards=Red_cards.tail(5)
fig = px.bar(Red_cards, y='Red cards',color=Red_cards.index)
fig.show()


Fouls=fwd_data[['Name','Fouls']]
Fouls=Fouls.sort_values(by=['Fouls'])
Fouls=Fouls.tail(5)
fig = px.bar(Fouls, x='Name',y='Fouls',color='Name')
fig.show()


Offsides=fwd_data[['Name','Offsides']]
Offsides=Offsides.sort_values(by=['Offsides'])
Offsides=Offsides.tail(5)
fig = px.bar(Offsides, x='Name',y='Offsides',color='Name')
fig.show()


Losses=fwd_data[['Name','Losses']]
Losses=Losses.sort_values(by=['Losses'])
Losses=Losses.tail(5)
fig = px.bar(Losses, x='Name',y='Losses',color='Name')
fig.show()


l6=list(fwd_data['Passes per match'])
for i in range(len(l6)):
    l6[i]=math.floor(l6[i])
fwd_data['Passes per match']=l6

Passes_per_match=fwd_data[['Name','Passes per match']]
Passes_per_match=Passes_per_match.sort_values(by=['Passes per match'])
Passes_per_match=Passes_per_match.tail(5)
fig = px.bar(Passes_per_match, x='Name',y='Passes per match',color='Name')
fig.show()


fwd_data['Shooting accuracy %']=fwd_data['Shooting accuracy %'].str.replace("%",'')
l=list(fwd_data['Shooting accuracy %'])
for i in range(len(l)):
    l[i]=int(l[i])
fwd_data['Shooting accuracy %']=l
Shooting_accuracy=fwd_data[['Name','Shooting accuracy %']]
Shooting_accuracy=Shooting_accuracy.sort_values(by=['Shooting accuracy %'])
Shooting_accuracy=Shooting_accuracy.tail(10)
fig = px.bar(Shooting_accuracy,x='Name', y='Shooting accuracy %',color='Name')
fig.show()


gk_data=gk_data.dropna(axis=0)


corr=gk_data.corr()
# Set up the matplotlib plot configuration
#
f, ax = plt.subplots(figsize=(20, 20))
#
# Generate a mask for upper traingle
#
mask = np.triu(np.ones_like(corr, dtype=bool))
#
# Configure a custom diverging colormap
#
cmap = sns.diverging_palette(230, 20, as_cmap=True)
#
# Draw the heatmap
#
sns.heatmap(corr, annot=True, mask = mask, cmap=cmap)

<AxesSubplot:>


wins=gk_data[['Name','Wins']]
wins=wins.sort_values(by=['Wins'])
wins=wins.tail(5)
fig = px.bar(wins,x='Name', y='Wins',color='Name')
fig.show()


Losses=gk_data[['Name','Losses']]
Losses=Losses.sort_values(by=['Losses'])
Losses=Losses.tail(5)
fig = px.bar(Losses,x='Name', y='Losses',color='Name')
fig.show()


gk_data['Saves']=gk_data['Saves'].str.replace(",",'')
l=list(gk_data['Saves'])
for i in range(len(l)):
    l[i]=int(l[i])
gk_data['Saves']=l
Saves=gk_data[['Name','Saves']]
Saves=Saves.sort_values(by=['Saves'])
Saves=Saves.tail(5)
fig = px.bar(Saves,x='Name', y='Saves',color='Name')
fig.show()


l=list(gk_data['Penalties Saved'])
for i in range(len(l)):
    l[i]=int(l[i])
gk_data['Penalties Saved']=l
Penalties_Saved=gk_data[['Name','Penalties Saved']]
Penalties_Saved=Penalties_Saved.sort_values(by=['Penalties Saved'])
Penalties_Saved=Penalties_Saved.tail(5)
fig = px.bar(Penalties_Saved,x='Name', y='Penalties Saved',color='Name')
fig.show()


l=list(gk_data['Catches'])
for i in range(len(l)):
    l[i]=int(l[i])
gk_data['Catches']=l
Catches=gk_data[['Name','Catches']]
Catches=Catches.sort_values(by=['Catches'])
Catches=Catches.tail(5)
fig = px.bar(Catches,x='Name', y='Catches',color='Name')
fig.show()


l=list(gk_data['Errors leading to goal'])
for i in range(len(l)):
    l[i]=int(l[i])
gk_data['Errors leading to goal']=l
Errors_goal=gk_data[['Name','Errors leading to goal']]
Errors_goal=Errors_goal.sort_values(by=['Errors leading to goal'])
Errors_goal=Errors_goal.tail(5)
fig = px.bar(Errors_goal,x='Name', y='Errors leading to goal',color='Name')
fig.show()


l=list(gk_data['Fouls'])
for i in range(len(l)):
    l[i]=int(l[i])
gk_data['Fouls']=l
Fouls=gk_data[['Name','Fouls']]
Fouls=Fouls.sort_values(by=['Fouls'])
Fouls=Fouls.tail(5)
fig = px.bar(Fouls,x='Name', y='Fouls',color='Name')
fig.show()


mid_data=mid_data.dropna(axis=0)


corr=mid_data.corr()
# Set up the matplotlib plot configuration
#
f, ax = plt.subplots(figsize=(20, 20))
#
# Generate a mask for upper traingle
#
mask = np.triu(np.ones_like(corr, dtype=bool))
#
# Configure a custom diverging colormap
#
cmap = sns.diverging_palette(230, 20, as_cmap=True)
#
# Draw the heatmap
#
sns.heatmap(corr, annot=True, mask = mask, cmap=cmap)

<AxesSubplot:>


Goals=pd.DataFrame(mid_data.groupby(['Name'])['Goals'].sum())
Goals=Goals.sort_values(by=['Goals'])
Goals=Goals.tail(5)
fig = px.bar(Goals, y='Goals',color=Goals.index)
fig.show()


Assists=pd.DataFrame(mid_data.groupby(['Name'])['Assists'].sum())
Assists=Assists.sort_values(by=['Assists'])
Assists=Assists.tail(5)
fig = px.bar(Assists, y='Assists',color=Goals.index)
fig.show()


Yellow_cards=pd.DataFrame(mid_data.groupby(['Name'])['Yellow cards'].sum())
Yellow_cards=Yellow_cards.sort_values(by=['Yellow cards'])
Yellow_cards=Yellow_cards.tail(5)
fig = px.bar(Yellow_cards, y='Yellow cards',color=Yellow_cards.index)
fig.show()


Red_cards=pd.DataFrame(mid_data.groupby(['Name'])['Red cards'].sum())
Red_cards=Red_cards.sort_values(by=['Red cards'])
Red_cards=Red_cards.tail(5)
fig = px.bar(Red_cards, y='Red cards',color=Red_cards.index)
fig.show()


Fouls=mid_data[['Name','Fouls']]
Fouls=Fouls.sort_values(by=['Fouls'])
Fouls=Fouls.tail(5)
fig = px.bar(Fouls,x='Name', y='Fouls',color='Name')
fig.show()


Offsides=mid_data[['Name','Offsides']]
Offsides=Offsides.sort_values(by=['Offsides'])
Offsides=Offsides.tail(5)
fig = px.bar(Offsides, x='Name',y='Offsides',color='Name')
fig.show()


mid_data['Duels won']=mid_data['Duels won'].str.replace(",",'')
l=[]
for i in mid_data['Duels won']:
    i=int(i)
    l.append(i)
mid_data['Duels won']=l


Duel_won=mid_data[['Name','Duels won']]
Duel_won=Duel_won.sort_values(by=['Duels won'])
Duel_won=Duel_won.tail(5)
fig = px.bar(Duel_won, x='Name',y='Duels won',color='Name')
fig.show()


mid_data['Duels lost']=mid_data['Duels lost'].str.replace(",",'')
l=[]
for i in mid_data['Duels lost']:
    i=int(i)
    l.append(i)
mid_data['Duels lost']=l


Duel_lost=mid_data[['Name','Duels lost']]
Duel_lost=Duel_lost.sort_values(by=['Duels lost'])
Duel_lost=Duel_lost.tail(5)
fig = px.bar(Duel_lost, x='Name',y='Duels lost',color='Name')
fig.show()


l6=list(mid_data['Passes per match'])
for i in range(len(l6)):
    l6[i]=math.floor(l6[i])
mid_data['Passes per match']=l6

Passes_per_match=mid_data[['Name','Passes per match']]
Passes_per_match=Passes_per_match.sort_values(by=['Passes per match'])
Passes_per_match=Passes_per_match.tail(5)
fig = px.bar(Passes_per_match, x='Name',y='Passes per match',color='Name')
fig.show()


l6=list(mid_data['Through balls'])
for i in range(len(l6)):
    l6[i]=int(l6[i])
mid_data['Through balls']=l6

Through_balls=mid_data[['Name','Through balls']]
Through_balls=Through_balls.sort_values(by=['Through balls'])
Through_balls=Through_balls.tail(5)
fig = px.bar(Through_balls, x='Name',y='Through balls',color='Name')
fig.show()


Interceptions=mid_data[['Name','Interceptions']]
Interceptions=Interceptions.sort_values(by=['Interceptions'])
Interceptions=Interceptions.tail(5)
fig = px.bar(Interceptions, x='Name',y='Interceptions',color='Name')
fig.show()


Passes_per_match=mid_data[['Name','Passes per match']]
Passes_per_match=Passes_per_match.sort_values(by=['Passes per match'])
Passes_per_match=Passes_per_match.tail(5)
fig = px.bar(Passes_per_match, x='Name',y='Passes per match',color='Name')
fig.show()


a=def_data.select_dtypes(include='object')


#cleaning data
for i in a.columns:
    def_data[i]=def_data[i].str.replace("%",'')
    def_data[i]=def_data[i].str.replace(",",'')


X = def_data.drop(['Goals Conceded','Unnamed: 0', 'Name'],axis=1)
y = def_data['Goals Conceded']


# Scaling
scaler = StandardScaler()
X = pd.DataFrame(scaler.fit_transform(X), columns=list(X.columns))
X


from sklearn import metrics
from sklearn.model_selection import cross_val_score



def print_evaluate(true, predicted):  
    mae = metrics.mean_absolute_error(true, predicted)
    mse = metrics.mean_squared_error(true, predicted)
    r2_score = metrics.r2_score(true, predicted)
    rmse = np.sqrt(metrics.mean_squared_error(true, predicted))

    print('MAE:', mae)
    print('MSE:', mse)
    print('RMSE:', rmse)
    print('r2_score:', r2_score)
    print('__________________________________')
    
def evaluate(true, predicted):
    mae = metrics.mean_absolute_error(true, predicted)
    mse = metrics.mean_squared_error(true, predicted)
    r2_score = metrics.r2_score(true, predicted)
    rmse = np.sqrt(metrics.mean_squared_error(true, predicted))
    return mae, mse, rmse,r2_score


from sklearn.model_selection import train_test_split

X_train, X_test, y_train, y_test = train_test_split(X, y, test_size=0.35, random_state=42)


from sklearn.linear_model import LinearRegression

lin_reg = LinearRegression(copy_X= True, fit_intercept= True, positive=False)
lin_reg.fit(X_train,y_train)
# print the intercept
print(lin_reg.intercept_)

63.889549667062965


#Coefficient terms of linear regression
coeff_df = pd.DataFrame(lin_reg.coef_, X.columns, columns=['Coefficient'])
coeff_df


#test data prediction
pred_linear = lin_reg.predict(X_test)
fig = px.scatter(x=y_test, y=pred_linear)
fig.show()


#calling the evaluation function that we created earlier for the test and train predictions
test_pred = lin_reg.predict(X_test)
train_pred = lin_reg.predict(X_train)

print('Test set evaluation:\n_____________________________________')
print_evaluate(y_test, test_pred)
print('Train set evaluation:\n_____________________________________')
print_evaluate(y_train, train_pred)

Test set evaluation:
_____________________________________
MAE: 5.7235258138258045
MSE: 105.47909700408378
RMSE: 10.270301699759544
r2_score: 0.9850276924111967
__________________________________
Train set evaluation:
_____________________________________
MAE: 4.048191842701582
MSE: 40.32298335407522
RMSE: 6.350038059261946
r2_score: 0.9955856420644812
__________________________________


#data frame of evaluations
results_df = pd.DataFrame(data=[["Linear Regression", *evaluate(y_test, test_pred) ]], 
                          columns=['Model', 'MAE', 'MSE', 'RMSE','R2square'])
results_df


from sklearn.ensemble import RandomForestRegressor

rf_reg = RandomForestRegressor()
rf_reg.fit(X_train, y_train)

test_pred = rf_reg.predict(X_test)
train_pred = rf_reg.predict(X_train)

print('Test set evaluation:\n_____________________________________')
print_evaluate(y_test, test_pred)

print('Train set evaluation:\n_____________________________________')
print_evaluate(y_train, train_pred)

Test set evaluation:
_____________________________________
MAE: 7.71360824742268
MSE: 255.95454020618553
RMSE: 15.998579318370288
r2_score: 0.9636683455436734
__________________________________
Train set evaluation:
_____________________________________
MAE: 3.22195530726257
MSE: 47.44123519553071
RMSE: 6.887759809657325
r2_score: 0.9948063715619138
__________________________________


#test data prediction
pred_rf = rf_reg.predict(X_test)
fig = px.scatter(x=y_test, y=pred_rf)
fig.show()


results_df_2 = pd.DataFrame(data=[["Random forest Regression", *evaluate(y_test, pred_rf) ]], 
                          columns=['Model', 'MAE', 'MSE', 'RMSE','R2square'])
results_df = results_df.append(results_df_2, ignore_index=True)
results_df


from sklearn.svm import SVR

svm_reg = SVR()
svm_reg.fit(X_train, y_train)

test_pred = svm_reg.predict(X_test)
train_pred = svm_reg.predict(X_train)

print('Test set evaluation:\n_____________________________________')
print_evaluate(y_test, test_pred)

print('Train set evaluation:\n_____________________________________')
print_evaluate(y_train, train_pred)

Test set evaluation:
_____________________________________
MAE: 37.698479974374706
MSE: 5591.15350771991
RMSE: 74.77401626046249
r2_score: 0.20635962506810457
__________________________________
Train set evaluation:
_____________________________________
MAE: 45.61397381914609
MSE: 8042.912438700849
RMSE: 89.68228609207533
r2_score: 0.11950229384816502
__________________________________


pred_svm = svm_reg.predict(X_test)
fig = px.scatter(x=y_test, y=pred_svm)
fig.show()


results_df_2 = pd.DataFrame(data=[["SVM Regression", *evaluate(y_test, pred_svm) ]], 
                          columns=['Model', 'MAE', 'MSE', 'RMSE','R2square'])
results_df = results_df.append(results_df_2, ignore_index=True)
results_df


from xgboost import XGBRegressor
xgb_reg = XGBRegressor()
xgb_reg.fit(X_train, y_train)

test_pred = xgb_reg.predict(X_test)
train_pred = xgb_reg.predict(X_train)

print('Test set evaluation:\n_____________________________________')
print_evaluate(y_test, test_pred)
print('====================================')
print('Train set evaluation:\n_____________________________________')
print_evaluate(y_train, train_pred)

Test set evaluation:
_____________________________________
MAE: 7.806387754840563
MSE: 267.73841627516344
RMSE: 16.362714208686878
r2_score: 0.9619956746344198
__________________________________
====================================
Train set evaluation:
_____________________________________
MAE: 0.00047246677846416596
MSE: 8.737051385146389e-07
RMSE: 0.0009347219578648182
r2_score: 0.9999999999043512
__________________________________


pred_xgb = xgb_reg.predict(X_test)
fig = px.scatter(x=y_test, y=pred_xgb)
fig.show()


from sklearn.tree import DecisionTreeRegressor
dt_reg = DecisionTreeRegressor()
dt_reg.fit(X_train, y_train)

test_pred = dt_reg.predict(X_test)
train_pred = dt_reg.predict(X_train)

print('Test set evaluation:\n_____________________________________')
print_evaluate(y_test, test_pred)
print('====================================')

Test set evaluation:
_____________________________________
MAE: 10.11340206185567
MSE: 429.9072164948454
RMSE: 20.734204023662095
r2_score: 0.9389765056506139
__________________________________
====================================


pred_dt= dt_reg.predict(X_test)
fig = px.scatter(x=y_test, y=pred_dt)
fig.show()


from sklearn.ensemble import GradientBoostingRegressor
gb_reg = GradientBoostingRegressor()
gb_reg.fit(X_train, y_train)

test_pred = gb_reg.predict(X_test)
train_pred = gb_reg.predict(X_train)

print('Test set evaluation:\n_____________________________________')
print_evaluate(y_test, test_pred)
print('====================================')
print('Train set evaluation:\n_____________________________________')
print_evaluate(y_train, train_pred)

Test set evaluation:
_____________________________________
MAE: 6.622537975198313
MSE: 220.4013951691796
RMSE: 14.845921836288227
r2_score: 0.9687149627253011
__________________________________
====================================
Train set evaluation:
_____________________________________
MAE: 0.5996057444963289
MSE: 0.7837504072318522
RMSE: 0.8852967904786802
r2_score: 0.999914198937136
__________________________________


pred_gb= gb_reg.predict(X_test)
fig = px.scatter(x=y_test, y=pred_gb)
fig.show()


results_df_2 = pd.DataFrame(data=[["Gradient boosting Regression", *evaluate(y_test, pred_gb) ]], 
                          columns=['Model', 'MAE', 'MSE', 'RMSE','R2square'])
results_df = results_df.append(results_df_2, ignore_index=True)
results_df


from sklearn.ensemble import BaggingRegressor
bagging_reg = BaggingRegressor()
bagging_reg.fit(X_train, y_train)

test_pred = bagging_reg.predict(X_test)
train_pred = bagging_reg.predict(X_train)

print('Test set evaluation:\n_____________________________________')
print_evaluate(y_test, test_pred)
print('====================================')
print('Train set evaluation:\n_____________________________________')
print_evaluate(y_train, train_pred)

Test set evaluation:
_____________________________________
MAE: 9.00721649484536
MSE: 396.0188659793815
RMSE: 19.900222762054234
r2_score: 0.9437868123559799
__________________________________
====================================
Train set evaluation:
_____________________________________
MAE: 4.10949720670391
MSE: 97.70759776536315
RMSE: 9.884715360867158
r2_score: 0.9893034623512693
__________________________________


pred_bagging= bagging_reg.predict(X_test)
fig = px.scatter(x=y_test, y=pred_bagging)
fig.show()


results_df_2 = pd.DataFrame(data=[["Bagging Regression", *evaluate(y_test, pred_bagging) ]], 
                          columns=['Model', 'MAE', 'MSE', 'RMSE','R2square'])
results_df = results_df.append(results_df_2, ignore_index=True)
results_df


results_df.set_index('Model', inplace=True)
results_df['MAE'].plot(kind='barh', figsize=(12, 8))

<AxesSubplot:ylabel='Model'>


results_df['R2square'].plot(kind='barh', figsize=(12, 8))

<AxesSubplot:ylabel='Model'>

	Unnamed: 0	Name	Appearances	Goals	Wins	Losses	Clean sheets	Goals Conceded	Tackles	Tackle success %	...	Goals with right foot	Goals with left foot	Hit woodwork	Big chances missed	Freekicks scored	Goals per match	Penalties scored	Shooting accuracy %	Shots	Shots on target
0	0	Sergio Reguilón	34.0	0.0	17.0	10.0	10.0	41.0	62.0	52%	...	0.0	0.0	0.0	NaN	NaN	NaN	NaN	NaN	NaN	NaN
1	1	Rhys Williams	9.0	0.0	7.0	1.0	3.0	6.0	6.0	67%	...	0.0	0.0	0.0	NaN	NaN	NaN	NaN	NaN	NaN	NaN
2	2	Emerson Royal	4.0	0.0	1.0	3.0	0.0	7.0	12.0	58%	...	0.0	0.0	0.0	NaN	NaN	NaN	NaN	NaN	NaN	NaN
3	3	Gary Cahill	394.0	28.0	172.0	135.0	103.0	491.0	484.0	75%	...	13.0	4.0	5.0	NaN	NaN	NaN	NaN	NaN	NaN	NaN
4	4	Mads Bech Sørensen	1.0	0.0	1.0	0.0	0.0	0.0	0.0	0%	...	0.0	0.0	0.0	NaN	NaN	NaN	NaN	NaN	NaN	NaN

	Unnamed: 0	Name	Appearances	Goals	Wins	Losses	Goals per match	Headed goals	Goals with right foot	Goals with left foot	...	Aerial battles lost	Aerial battles won	Cross accuracy %	Duels lost	Duels won	Errors leading to goal	Recoveries	Successful 50/50s	Tackle success %	Through balls
0	0	Emmanuel Longelo	0	0	0	0	0.00	0	0	0	...	NaN	NaN	NaN	NaN	NaN	NaN	NaN	NaN	NaN	NaN
1	1	Ayoze Pérez	203	43	69	88	0.21	6	31	6	...	NaN	NaN	NaN	NaN	NaN	NaN	NaN	NaN	NaN	NaN
2	2	Sadio Mané	236	99	142	42	0.42	12	60	27	...	NaN	NaN	NaN	NaN	NaN	NaN	NaN	NaN	NaN	NaN
3	3	Raúl Jiménez	93	35	39	29	0.38	9	22	4	...	NaN	NaN	NaN	NaN	NaN	NaN	NaN	NaN	NaN	NaN
4	4	Dwight Gayle	139	26	32	74	0.19	3	16	7	...	NaN	NaN	NaN	NaN	NaN	NaN	NaN	NaN	NaN	NaN

	Appearances	Goals	Wins	Losses	Clean sheets	Tackles	Tackle success %	Last man tackles	Blocked shots	Interceptions	...	Through balls	Accurate long balls	Yellow cards	Red cards	Fouls	Offsides	Headed goals	Goals with right foot	Goals with left foot	Hit woodwork
0	-0.256438	-0.516374	-0.138744	-0.332310	-0.158238	-0.197278	0.310105	-0.413519	0.084628	-0.230457	...	-0.203931	-0.401542	-0.213296	-0.446520	-0.189627	-0.293683	-0.426606	-0.426264	-0.313002	-0.458332
1	-0.588723	-0.516374	-0.446074	-0.653625	-0.504855	-0.614223	0.780546	-0.413519	-0.401982	-0.605535	...	-0.203931	-0.509068	-0.632595	-0.446520	-0.602982	-0.519468	-0.426606	-0.426264	-0.313002	-0.458332
2	-0.655180	-0.516374	-0.630472	-0.582222	-0.653406	-0.569550	0.498281	-0.413519	-0.483083	-0.605535	...	-0.354023	-0.588543	-0.632595	-0.446520	-0.587084	-0.519468	-0.426606	-0.426264	-0.313002	-0.458332
3	4.528473	6.258730	4.624870	4.130400	4.446818	2.944705	1.031448	8.403491	4.788518	3.398201	...	0.546534	3.586238	2.805660	4.117906	3.753144	0.835238	4.084626	8.118070	2.086680	2.340333
4	-0.695054	-0.516374	-0.630472	-0.689327	-0.653406	-0.658896	-1.320758	-0.413519	-0.564185	-0.657871	...	-0.354023	-0.597893	-0.632595	-0.446520	-0.666575	-0.519468	-0.426606	-0.426264	-0.313002	-0.458332
...	...	...	...	...	...	...	...	...	...	...	...	...	...	...	...	...	...	...	...	...	...
271	0.248636	0.693466	0.322251	0.096110	-0.009688	-0.063259	0.561007	0.076315	0.165729	-0.056002	...	-0.203931	0.383859	0.122144	-0.446520	0.223728	0.383670	1.377887	0.230992	-0.313002	-0.458332
272	-0.708346	-0.516374	-0.661205	-0.689327	-0.653406	-0.658896	-1.320758	-0.413519	-0.564185	-0.657871	...	-0.354023	-0.597893	-0.632595	-0.446520	-0.666575	-0.519468	-0.426606	-0.426264	-0.313002	-0.458332
273	-0.708346	-0.516374	-0.661205	-0.689327	-0.653406	-0.658896	-1.320758	-0.413519	-0.564185	-0.657871	...	-0.354023	-0.597893	-0.632595	-0.446520	-0.666575	-0.519468	-0.426606	-0.426264	-0.313002	-0.458332
274	-0.189981	-0.516374	-0.384608	-0.010995	-0.554372	0.078204	0.811909	-0.413519	0.003526	-0.300239	...	-0.354023	-0.424917	-0.213296	-0.446520	-0.332712	0.835238	-0.426606	-0.426264	-0.313002	-0.458332
275	0.102431	-0.032438	0.107120	0.167513	-0.207755	-0.055814	0.623732	-0.413519	-0.239778	-0.090893	...	-0.354023	-0.139742	0.038284	1.074955	0.048847	-0.293683	0.475641	-0.426264	-0.313002	-0.458332

	Unnamed: 0	Name	Appearances	Clean sheets	Wins	Losses	Saves	Penalties Saved	Punches	High Claims	...	Errors leading to goal	Own goals	Yellow cards	Red cards	Fouls	Assists	Passes	Passes per match	Accurate long balls
0	0	Scott Carson	146.0	30.0	45.0	66.0	435	4.0	66.0	161.0	...	10.0	1.0	6.0	1.0	4.0	2.0	2,732	18.71	832
1	1	James Trafford	0.0	0.0	0.0	0.0	0	0.0	0.0	0.0	...	0.0	0.0	0.0	0.0	0.0	0.0	0	0.00	0
2	2	John Ruddy	121.0	25.0	33.0	56.0	392	2.0	51.0	177.0	...	12.0	0.0	3.0	1.0	1.0	0.0	2,594	21.44	739
3	3	Nathan Shepperd	0.0	0.0	0.0	0.0	0	0.0	0.0	0.0	...	0.0	0.0	0.0	0.0	0.0	0.0	0	0.00	0
4	4	Illan Meslier	42.0	12.0	17.0	17.0	168	1.0	29.0	31.0	...	2.0	1.0	1.0	0.0	1.0	0.0	1,315	31.31	190

	Unnamed: 0	Name	Appearances	Goals	Wins	Losses	Goals per match	Headed goals	Goals with right foot	Goals with left foot	...	Interceptions	Clearances	Headed Clearance	Recoveries	Duels won	Duels lost	Successful 50/50s	Aerial battles won	Aerial battles lost	Errors leading to goal
0	0	Juan Mata	271	52	143	62	0.19	2	7	43	...	135	66	6	1,109	624	907	175.0	23.0	70.0	1.0
1	1	Ethan Galbraith	0	0	0	0	0.00	0	0	0	...	0	0	0	0	0	0	0.0	0.0	0.0	0.0
2	2	Lukas Rupp	17	0	2	13	0.00	0	0	0	...	18	6	3	70	39	71	6.0	6.0	17.0	0.0
3	3	Imran Louza	1	0	0	1	0.00	0	0	0	...	1	1	1	2	3	2	1.0	1.0	1.0	0.0
4	4	Tom Dele-Bashiru	0	0	0	0	0.00	0	0	0	...	0	0	0	0	0	0	0.0	0.0	0.0	0.0

	Coefficient
Appearances	86.583676
Goals	59.563199
Wins	-24.161112
Losses	31.067559
Clean sheets	-13.628617
Tackles	-12.376671
Tackle success %	-1.221890
Last man tackles	-3.367268
Blocked shots	4.930065
Interceptions	9.463306
Clearances	-9.682706
Headed Clearance	18.845830
Clearances off line	2.140934
Recoveries	0.750020
Duels won	23.745986
Duels lost	-9.585262
Successful 50/50s	-3.534413
Aerial battles won	-13.308394
Aerial battles lost	-2.907337
Own goals	-0.278132
Errors leading to goal	0.839827
Assists	7.547084
Passes	13.898412
Passes per match	0.496066
Big Chances Created	2.825995
Crosses	-8.753661
Cross accuracy %	0.526406
Through balls	2.376542
Accurate long balls	-11.773216
Yellow cards	-0.093989
Red cards	-2.469427
Fouls	-1.741448
Offsides	-3.418131
Headed goals	-33.875491
Goals with right foot	-20.212991
Goals with left foot	-22.262241
Hit woodwork	2.564262

	Model	MAE	MSE	RMSE	R2square
0	Linear Regression	5.723526	105.479097	10.270302	0.985028
1	Random forest Regression	7.713608	255.954540	15.998579	0.963668

Data Analysis¶

Defenders data¶

top 5 Goals¶

top 5 players most Yellow cards¶

top 5 players most Red cards¶

top 5 players most Tackle success %¶

top 5 players most Duels won¶

top 5 players most Duels lost¶

top 5 players most Aerial battles won¶

top 5 players most Aerial battles lost¶

top 5 players most Errors leading to goal¶

top 5 players most Passes per match¶

top 5 players most Duels won¶

top 5 players most Duels lost¶

top 5 players most Cross accuracy %¶

fwd_data¶

top 5 players most Goals¶

top 5 players most Yellow cards¶

top 5 players most Red cards¶

top 5 players most Fouls¶

top 5 players most Offsides¶

top 5 players most Losses¶

top 5 players most Passes per match¶

top 5 players most Shooting accuracy %¶

gk data¶

top 5 players most Wins¶

top 5 players most Losses¶

top 5 players most Saves¶

top 5 players most Penalties Saved¶

top 5 players most Catches¶

top 5 players most Errors leading to goal¶

top 5 players most Fouls¶

mid data set :¶

top 5 players most Goals¶

top 5 players most Assists¶

top 5 players most Yellow cards¶

top 5 players most Red cards¶

top 5 players most Fouls¶

top 5 players most Offsides¶

top 5 players most Duels won¶

top 5 players most Duels lost¶

top 5 players most Passes per match¶

top 5 players most Through balls¶

top 5 players most Interceptions¶

top 5 players most Passes per match'¶

Data preprocessing:¶

defining inputs and output as X and y respectively¶

Scaling input features¶

splitting the dataset into testing and training respectively¶

Modeling¶

Actual vs predicted values for linear regression¶

test set evaluations¶

Model 2:RandomForestRegressor¶

Actual vs predicted values for RandomForestRegressor¶

Model 3:SVR¶

Actual vs predicted values for SVR Regressor¶

Model 4:XGBRegressor¶

Actual vs predicted values for XGBRegressor¶

Model 5:DecisionTreeRegressor¶

Model 6:GradientBoostingRegressor¶

Model 7:BaggingRegressor¶

Evaluation for all the models on test data¶

R2square of all the models¶

Conclustion:¶